International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 184

ISSN 2229-5518

QoS-based Web Service Selection Using

Filtering, Ranking and Selection Algorithm

Tajudeen Adeyemi Ajao, Safaai Deris, Isiaka Adekunle Obasa

Abstract— Appropriate web service selection procedure for determining optimal web service for a requester still remain an active area of research owing to the persisting upward trends in services with similar functionality. This paper proposes a QoS-based Filtering, Ranking and Selection Algorithm for the purpose of selecting the best web service for requester in line with his/her preferences. Experiments are conducted using real web services datasets and the outcome of the experiments confirms an improvement over existing methods.

Index Terms— Quality of Service, Service Filtering, Service Ranking, Web service, Web Service Selection

——————————  ——————————

1 INTRODUCTION

he basic building block of Service Oriented Architecture (SOA) is the web service. In web service architecture, ser- vice provider presents web services that offers tasks or

business procedures which are set up over the internet, for invocation by clients; a web service requester defines require- ments for the purpose of finding web services of interest. Pub- lishing, binding, and discovering web services are the three major tasks in web service architecture. The web service archi- tecture in Figure 1 illustrates the service requester, providers, and discovery system with their interactions.
Figure 1: Web Services Architecture (Govatos, 2002)
As shown in figure 1, the service providers build web services that offer specified functions for users’ which is made availa-

————————————————

• Tajudeen Adeyemi Ajao is currently pursuing masters degree program in computer science in Universiti Teknologi Malaysia, Malaysia, PH-

601116297202. E-mail: taajao2@live.utm.my

• Safaai Deris is a professor of Computer Science in Universiti Teknologi

Malaysia, Malaysia, PH-6075533033. E-mail: safaai@utm.my

• Isiaka Adekunle Obasa is currently pursuing PhD program in computer

science in Universiti Teknologi Malaysia, Malaysia, PH-60146164635.E-

mail: aiobasa2@live.utm.my

ble on the internet for their consumption. The web service re- quester is any user of the web service who describes and sub mits requests for the purpose of finding a service. The web service registry is a centralized directory of services where service providers publish their service information. The speci- fied information is kept in the registry and examined on sub- mission of request by requester. Universal Description, Dis- covery and Integration (UDDI) is the registry standard for Web services.

1.1 Web Service

Varieties of definitions of web services are given by re- searchers and web service consortia. According to World Wide Web Consortium [1], “A Web service is a software sys- tem designed to support interoperable machine-to-machine interaction over a network. It has an interface described in a machine-processable format (specifically WSDL). Other sys- tems interact with the Web service in a manner prescribed by its description using SOAP messages, typically conveyed us- ing HTTP with an XML serialization in conjunction with other Web-related standards.” Web Services are self-independent application that exhibits modular and distributed concepts [2]. Web services can be used by any application irrespective of platform in which it is developed. Web service description is provided in WSDL document, it can be accessed from internet using SOAP protocol. The primary aim of Web services is to demystify and normalize application interoperability within and across establishments, leading to growth in operational proficiencies and intimate partner relations [3]. In industry, many applications are built by calling different web services available on the internet which results in overwhelming ac- ceptance of web services in recent years and the trend will continue for many years to come.

1.2 QoS Requirements for Web Services

According to W3C [4], Quality of Service (QoS) denotes the quality aspects of a web service such as performance, reliabil- ity, scalability, availability, etc. Constraints are defined on the QoS, and these constraints can be utilized to select an optimal service for a requester. For example, a requester can request for weather information service with availability of 96%. QoS plays vital role in all service oriented tasks, particularly in the discovery and selection of optimal services. In a situation

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 185

ISSN 2229-5518

where multiple services providing similar functionality can accomplish a user’s functional requirement, QoS provides a reliable means of differentiating between the services, hence, QoS is an essential factor for choosing optimal service for re- questers.

1.3 Web Service Selection

Web service selection refers to the process by which a service implementation is chosen from numerous services discovered in response to requester’s functional requirement. Service discovery is a prerequisite for service selection process; however, service selection is a core issue that must be addressed in order to retrieve appropriate service for a requester. Functional and Non-Functional properties especially QoS are the two main classes of requirements that are considered in selecting optimal service for a requester. Much work has been done in the domain of web service discovery, which mainly focusses on functional properties of web services. However, in view of large number of services with comparable functionalities, web service discovery alone is inadequate for selecting optimal service that would satisfy users’ expectations, hence; efficient methodologies and procedures are required for appropriate web service selection, which is the main concern in the domain of service oriented computing.

1.4 Background of the Study

QoS related approaches for optimal selection of web ser- vices have been discussed in a number of recent literatures. The existing works examined various methods by which op- timal web services can be identified from a set of candidates offering similar functionality using the QoS performance of the candidates and the preference of web service requesters. However, previous approaches failed in one way or the other in considering preferences of requesters which led to recom- mending the same service with highest score of QoS points to different requesters in spite of their diverse QoS inclinations. Some researchers arbitrarily assign unrealistic weights of zero to the weight of parameters not specified by the requester thereby putting the quality of output of such experiment into doubt. In addition, computation of normalized QoS by some of the existing approaches is debatable. Some expects users to specify constraints as well as assign weight to each constraint parameter which results in unnecessary burden on the re- quester and prone to conflicting representation.
An effective approach should recognize users’ preferences and recommend appropriate service to each requester in line with his/her QoS inclination. This research work, attempts to ad- dress the above mentioned issues by using QoS-based service filtering, ranking and selection algorithm as a means of select- ing best service for the requester with consideration of user’s preferences and also deriving weights from user’s specified constraints, thereby relieving users of the burden of providing weights for QoS parameters.

2 LITERATURE REVIEW

Service filtering is one of the methods used in managing nu- merous services discovered in response to requester’s query.
The aim of service filtering is to expose requesters to only those services that are relevant to their requests while blocking non-valuable services from requester’s view. A number of approaches are proposed in the literature for web service se- lection using filtering techniques on QoS properties of the ser- vices. Some of the approaches in this area include reputation- aware/user ratings, Context-based and Knowledge-based. Reputation-aware selection of web service is a method of se- lecting a web service using reputation which emanates from the public’s opinion about the quality of a web service (such as availability, response time, reliability etc). It is unbiased and denotes a combined evaluation of a collection of individuals. Reputation mechanism utilizes consumers’ responses to win- now relevant services from those services that are redundant [5]. A number of trust and reputation techniques have been offered for web service selection. Most of these techniques depend on central QoS registry for the collection and storage of feedbacks from consumers. In these techniques, consumers send the data acquired from operating a web service (e.g. reli- ability, availability, response time values) to the central QoS registry in line with the QoS information and consumer’s pro- file that shows the consumer’s ratings for various QoS metrics (i.e. the importance attached to the QoS metrics by a consum- er), the QoS registry computes an overall rating for each web service that agrees with the consumer’s search request. The web service with the highest score is then recommended to the consumer.
Context-aware services are those services that have their de- scription enriched with context information in line with the service execution environment [6]. Context-aware systems have been used in recommendation system and it aims at en- hancing the quality of recommendations by taking into con- sideration available contextual information, such as location, time, mood, or presence of others. Context-based filtering ap- proach explores users past consumption pattern and recom- mend based on such interest. The challenges of context based approach include temporal, mood, and social dimensions of context [7].
Knowledge-based filtering technique involves user’s specifica- tion of his requirements. Recommendation of items is done in line with the user’s specified constraints. The approach re- quires requester to submit requirements using predefined in- terfaces. This information is then compared with the descrip- tions of the available services to identify potential service can- didates. The services suitable to requester are displayed in ranked order [8].
Some of the works done on web service selection with inclina- tion on QoS-aware are given below:
Yu et al. [9] proposed two models for the QoS-based service composition problem: combinatorial and graph model. They used two heuristic algorithms based on linear programming for service procedures with a serial flow and a general flow structure to enable the selection of QoS-oriented services. The problem of the algorithms lies in scalability.
Xu et al. [10] combined reputation-enhanced service dis- covery algorithm, which utilizes consumer’s feedbacks for QoS computation and a reputation management system used for building and maintaining service reputations for web ser-

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 186

ISSN 2229-5518

vice discovery and selection. Their algorithm discovers a set of services that match the consumer’s requests. It also ranks the services using the QoS computed and reputation scores based on consumer’s preferences in the service discovery request. However, the authenticity of the advertised QoS information presented may be doubtful.
Al-Masri & Mahmoud [11] proposed a solution by intro- ducing the term -Web Service Relevancy Function (WsRF) which is used to measure the relevancy ranking of a specific Web service using QoS parameters and preference of requester in order to determine the best obtainable Web service during Web services discovery process based on a set of preferences specified by requester. The study though, recognizes user’s preferences but, it places additional burden on user to specify weights for QoS parameters. Also the use of cost as one of the QoS parameters is subjected to argument as cost was excluded from QoS parameters specified by W3C.
D'Mello et al. [12] present a QoS broker based architecture for the purpose of selecting Web service dynamically. They compute QoS score using Quality Constraints Tree mechanism (QCT) for functionally similar Web Service. Min-Max normali- zation method was used for determining best Web Service for requester in response to his/her QoS requirements specifica- tion along with functional requirements. Also, their approach relies on user to supply weight for each QoS parameter which is a burden.
Zheng et al. [13] proposed a Web service recommender sys- tem (WSRec) which incorporates user-contribution machinery for Web service QoS information gathering with a hybrid col- lective filtering algorithm. They propose Web service QoS value forecast which can be used for Web service recommen- dation and selection. Their approach involves complex com- putations.
Malik & Bouguettaya [14] propose a reputation manage- ment framework for the establishment of trust between web services. Their framework made a provision for a cooperative model where sharing of experiences is established between web services regarding their service providers with their peers through response ratings. Various ratings are aggregated to derive service provider’s reputation used in evaluating trust. A set of techniques was devised which aim at precisely accu- mulating the submitted ratings for reputation valuation.
Raj & Sasipraba [15] proposed web service selection model for selecting best web service based on QoS constraints. They stored the QoS attributes of web services in a database. The user specifies functional requirements, QoS values and threshold value for response time and throughput which are used for filtering the related service from the list. Min-Max method is used for normalizing QoS values. This approach places additional burden on the user for having to specify threshold aside from QoS constraints for his preferences. Also, the use of threshold for computation of normalized data for negatively inclined parameters e.g. response time is debatable.
Li et al. [16] present a selection approach tagged fast web service selection (FWSS) for service composition system. They used hierarchical fuzzy system and mixed integer program- ming to locate the most optimal service for the requester. This approach requires high computation which is a drawback.
Meng et al. [17] present requester’s preferences obtained from past QoS values. They propose a QoS model in which users are allowed to specify their preferences while providing combination of multiple QoS properties to give an overall rat- ing to a service. Then, the similarities between users are measured using the correlation between their rankings of ser- vices.
Maheswari & Karpagam [18] proposed a framework for QoS based Semantic Web Service Selection. The framework consists of four components -OWL-S converter, Semantic Re- pository, QoS Broker and Matchmaker. The framework de- termines the Web Service Relevance Function (WsRF) using the normalization process and then selects the relevant web services for requester. Their approach is dependent on per- ceived QoS by the users which in some cases may not be relia- ble.

3 METHODOLOGY

The proposed system is designed to carry out the process of selecting optimal service for a requester using service filtering, ranking and selection algorithm. The following four attributes
–Response time, Reliability, Availability and Successability are used in this research work. Explanation of these attributes is given in table 1.
Table 1: Explanation on QoS Parameters used in Proposed
Method
The four parameters are chosen because of their relative im- pact in assisting requesters to make reasonable selection deci- sion as they relate to dependability and performance metrics [19] which are fundamental qualities of web services that are necessary for the fulfillment of web service requester’s objec- tives. Dependability relates to building confidence about a web service. Reliability, availability and successability come under dependability metrics. The performance of a web ser- vice represents how fast a service request can be completed. Response time is categorized under performance. Reliability, availability and successability are considered crucial attributes

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 187

ISSN 2229-5518

in service selection; the response time also becomes very im- portant since majority of consumers expects low response time in retrieving services of interest [20].

3.1 Service Filtering, Ranking and Selection

qmax != qmin, 1 if qmax = qmin
……(1)
……(2)

The service filtering, ranking and selection algorithm is con- cerned with filtering out redundant services, normalizing the QoS values of each parameter, computing overall QoS score for each service, arranging the relevant services in descending order of overall QoS scores and recommending the best ser- vice (service with the highest score) to the requester based on his preferences. Figure 2 depicts this algorithm.
/*input:
a set of n candidates s(f) = (s1,s2 ,s3 ,…,sn) that each fulfills requester’s functional requirements; //this is output from discovery agent
a set of thresholds (default constraints) for desired services having 4 elements t = <t1, t2 , t 3, t 4>; //where t 1 , t2 , t 3, t 4 = response time, availability, successability and reliability values.
a set of constraints for desired services having 4 elements
c = <c1, c2 , c3 , c4 >; //where c1, c2, c3, c4 = response time, reliability, availability and successability values*/
/*output:
an optimal service sp ∈ sf that fulfills requester’s function-
al and nonfunctional requirements*/
// initialization:
1 enter threshold data // (default values to be used if user did
// not enter data for constraint parameter(s))
2 store threshold data
// accept qos constraints
3 user enters constraints requirements
//filtering: compare each candidate’s qos value with user’s
//constraints
4 calculate total no of candidates (n)
5 while i < = n do
6 for j = 1 to 4
7 if q ij (s i) < cj then filter out service s i // (filter out
// the current candidate web service )
8 endif
9 end for
10 end while
//ranking:
11 compute normalized qos data for each filtered service
12 compute weighted values for the constraints //based on re
// quester’s requirement
13 compute product of weight and normalized QoS values for each service and get total scores for each service
14 sort the services in descending order based on scores computed
15 return the first service in the list
Figure 2: Filtering, Ranking and Selection Algorithm

3.2 Normalization of QoS values

In the normalization process, equation 1 is used for reliability availability and successability parameters that require maxi- mization whereas equation 2 is used for response time that requires minimization [21].
qp, qn represent normalized value for positively and nega- tively inclined QoS parameter respectively, qmax and qmin represent the maximum and minimum QoS values for a set of QoS parameters and q is the QoS value of the parameter being considered.
The QoS values for the constraints are normalized using the following formula:
……(3) ……(4)
qmax != y, 1 if qmax = y
where qc represents the QoS value for the parameter being considered, y is the default threshold value for the parameter being considered. q and q’ depict normalized value for posi- tively and negatively inclined parameter respectively.

3.3 Service Selection Process

When a service requester submits his query for a service of interest for example getting a map from global positioning system (gps), the Web service discovery agent returns those services that meet the requester’s functional requirement. Be- fore service selection is done, the consumer need to specify the constraints for the selection. As an example, the requester asks for a weather service satisfying these constraints: response time less than 350ms, availability not less than 85%, successa- bility greater than 80%, reliability greater than 70%.
Based on Service Filtering, Ranking and Selection Algorithm, those services that fail to satisfy the requester’s specified con- straints are winnowed out. The QoS values of the filtered can- didates are normalized using Min-Max method. The total QoS scores is computed and used for ranking the candidates. final- ly, the top n (n <=5) filtered candidates are arranged in the order of significance and presented to the requester with em- phasis on the service with the highest score in the ranked list as the recommended optimal service.

4 EXPERIMENT AND RESULTS

Experiments are conducted on the proposed QoS-based Filter- ing Ranking and Selection Algorithm (QFRSA) and the output compared with the output from Web Service Selection and Ranking Model (WSSRM) proposed by Raj and Sasipraba (2010) which also used max-min method of normalization but computation for minimizing negatively inclined parameters is done using equation 5.

…..(5)

Three Requesters (A, B and C) were considered with each of them having the same functional requirement but different

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 188

ISSN 2229-5518

QoS constraints. The functional requirement for each of the users is email verifier.
The list of services returned by discovery agent satisfying the functional requirement (email verifier) for all the three re- questers is presented in Table 2 below.
Table 2: List of functional candidates for requesters A, B & C
with their QoS values
Table 5: Ranked weighted Sum of Normalized Filtered Candidates for Requester A, returned by WSSRM
S1 having the highest score (0.25056 in QFRSA and 0.25089 in WSSRM) in the lists is recommended as the best service for requester A by both methods.
The QoS constraints presented by each of the requesters are shown in Table 3.

Table 3: QoS constraints for Requesters A, B and C

4.2 Test Scenario 2: Requester B

For requester B with constraints (550,80,80,60), the ranked fil- tered candidate services with the scores in response to his que- ry are shown in Table 6 and Table 7 for QFRSA and WSSRM respectively.
Table 6: Ranked weighted Sum of Normalized Filtered Candi- dates for Requester B returned by QFRSA

The requesters specify their QoS parameters for response time, availability, successability and reliability in that order and each of the users’ QoS constraints are utilized in turn on the two models. Requester C is indifferent regarding the reliability parameter; in this case, the default value of 50 is used in carry- ing out the test. The subsititution of 0 for the default value is necessary in assisting the requester to select a suitable service owing tio the fact that, selecting a service with negligible reli- ability is unrealistic. The simulations are presented in scenari- os 1 – 3.

4.1 Test Scenario 1: Requester A

Based on constraints (600,60,60,66) given by requester A, the ranked filtered candidates in response to his query with the overall score for each service is shown in Table 4 and 5 for QFRSA and WSSRM respectively.
Table 4: Ranked weighted Sum of Normalized Filtered Candi- dates for Requester A, returned by QFRSA
Table 7: Ranked weighted Sum of Normalized Filtered Candi- dates for Requester B returned by WSSRM

S7 having the highest score (0.41522) in the list is recommend- ed as the best service for requester B from both QFRSA and WSSRM.

4.3 Test Scenario 3: Requester C

For requester C with constraints (500,80,65,50), (a default value of 50 is used to substitute 0 since requester C is indiffer- ent to reliability parameter) the ranked filtered candidate ser- vices in response to his query are shown in Tables 8 and 9 for QFRSA and WSSRM.

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 189

ISSN 2229-5518

Table 8: Ranked weighted Sum of Normalized Filtered Candi- dates for Requester C returned by QFRSA
Table 11: Normalized QoS values for S1, S5, S7 (optimal Qos value) and Euclidean distance value (Case of Requester B)

Table 9: Ranked weighted Sum of Normalized Filtered Candi- dates for Requester C returned by WSSRM

4.4 Discussion

In the following section, summary of the results generated from the models are presented for better evaluation. The web service selection and ranking model (WSSRM) proposed by Raj and Sasipraba (2010) is used as baseline. The results of ranking by the proposed method (QFRSA) and the WSSRM proposed by Raj & Sasipraba are presented in table 10.
Table 10: Comparison of outputs from existing and proposed models
From table 11, euclidean distances of S1 and S5 from the opti- mal service returned for requester B are 1.291 and 0.528 re- spectively. This result shows that service S5 is preferred to service S1 which agrees with the output from QFRSA.
Also for requester C, in the list generated by QFRSA, service S1 has higher ranking position than S6 whereas they appeared in opposite order in the list generated from WSSRM. A further scrutiny of QoS values of the parameters for services S1 and S6 using Euclidean distance indicates that S6 has closer similarity to the optimal service returned by the two models as depicted in table 12.
Table 12: Normalized QoS values for S1, S6, S7 (optimal Qos value) and Euclidean distance value (Case of Requester C)
Euclidean distance of S1 and S6 from the optimal service re- turned for requester C gives 0.481 and 0.56522 respectively. This confirms that service S6 is preferred to service S1 and this agrees with the output from QFRSA. The cases illustrated above clearly indicate that the QFRSA approach performs bet- ter than the WSSRM proposed by Raj and Sasipraba.
WSSRM and QFRSA generate similar ranking list of services for requester A and the scores for the services are almost at par for all services except service S4 whose score is slightly higher in WSSRM than that of QFRSA. However, for re- quester B, there is an exclusion of S5 in the rank list from WSSRM which was ranked higher than S1 in the list from QFRSA. A close examination of QoS values of all parameters for services S1 and S5 using Euclidean distance shows that S5 has closer similarity to the optimal service returned by both models as shown in table 11.

5 CONCLUSION

In this paper QoS-based Filtering and Ranking Algorithm is proposed for selecting best web service for requesters. The procedure allows users to specify their QoS constraints which are used to filter off redundant services and compute total QoS utility score for each filtered candidate. It filters off re- dundant services, ranks relevant web services and assists us- ers in selecting the best web service in response to their speci- fied preferences.
Based on this research work, it can be concluded that, the pro- posed QFRSA provides solution to dynamic web service selec- tion at run time. The output from the proposed QFRSA demonstrates that the probability of selecting a service that best meet user’s requirements is improved if the user specifies QoS constraints, and that user’s specification of his/her pref-

IJSER © 2013 http://www.ijser.org

International Journal of Scientific & Engineering Research, Volume 4, Issue 7, July-2013 190

ISSN 2229-5518

erences is a key towards selecting optimal service for his/her request. Also, QFRSA produced better quality result in com- parison to the existing methods. The future scope of this work is to include additional QoS parameters.

REFERENCES

[1] W3C Working Group Web Services Architecture. Available at http://www.w3c.org/TR/ws-arch/ [Accessed on

05/11/2012].

[2] IBM WSAT, Web Service Architecture. Available at: http://www.ibm.com/developerworks/webservices/library/w- ovr/ [Accessed on 23/10/2012].

[3] Govatos, G. UDDI is Yellow Pages of Web services. Network

World Web Applications Newsletter. Available at http://www.networkworld.com/news/tech/2002/0527tech.html [Accessed on 21/12/2012].

[4] W3C Working Group, QoS for Web Services:Requirements

and Possible Approaches. [Available at http://www.w3c.or.kr/kr-office/TR/2003/ws-qos/ [Accessed on

15/11/2012]

[5] Wang, Y. and J. Vassileva. A review on trust and reputation for web service selection. in Distributed Computing Systems Workshops, 2007. ICDCSW'07. 27th International Conference. 2007. IEEE.

[6] Kirsch-Pinheiro, M., Y. Vanrompay, and Y. Berbers.

Context-aware service selection using graph matching. in 2nd Non Functional Properties and Service Level Agreements in Service Oriented Computing Workshop (NFPSLA-SOC'08), ECOWS. CEUR Workshop proceedings. 2008.

[7] Said, A., S. Berkovsky, and E.W. De Luca. Putting things

in context: Challenge on context-aware movie recommendation. in Proceedings of the Workshop on Context- Aware Movie Recommendation. 2010. ACM.

[8] Godse, M., R. Sonar, and A. Jadhav, A hybrid approach for

knowledge-based product recommendation, in Information

Systems, Technology and Management. 2009, Springer. p.

268-279.

[9] Yu, T., Y. Zhang, and K. Lin, Efficient algorithms for Web services selection with end-to-end QoS constraints. ACM Transactions on the Web (TWEB), 2007. 1(1): p. 6.

[10] Xu, Z., P. Martin, W. Powley, and F. Zulkernine.

Reputation-enhanced qos-based web services discovery. in Web

Services, 2007. ICWS 2007. IEEE International Conference.

2007. IEEE.

[11] Al-Masri, E. and Q.H. Mahmoud. Discovering the best web service. in Proceedings of the 16th international conference on World Wide Web. 2007. ACM.

[12] D'Mello, D.A., V.S. Ananthanarayana, and T. Santhi. A qos broker based architecture for dynamic web service selection. in Modeling & Simulation, 2008. AICMS 08. Second Asia International Conference. 2008. IEEE.

[13] Zheng, Z., H. Ma, M.R. Lyu, and I. King. Wsrec: A

collaborative filtering based web service recommender system.

in Web Services, 2009. ICWS 2009. IEEE International

Conference. 2009. IEEE.

[14] Malik, Z. and A. Bouguettaya, Rateweb: Reputation assessment for trust establishment among web services. The VLDB Journal, 2009. 18(4): p. 885-911.

[15] Raj, R.J. and T. Sasipraba. Web service selection based on QoS Constraints. in Trendz in Information Sciences & Computing (TISC). 2010. IEEE.

[16] Li, J., S. Wang, Q. Sun, and F. Yang, A Fast Web Service

Selection Approach. Computing and Intelligent Systems,

2011: p. 502-507.

[17] Meng, L., J. Li, and H. Sun. WSRank: A Collaborative Ranking Approach for Web Service Selection. in Computer and Information Technology (CIT), 2011 IEEE 11th International Conference. 2011. IEEE.

[18] Maheswari, S. and G.R. Karpagam, QoS Based Efficient

Web Service Selection. European Journal of Scientific

Research, 2011. 66(3): p. 428-440.

[19] Marzolla, M. and R. Mirandola, Qos analysis for web service applications: a survey of performance-oriented approaches from an architectural viewpoint, 2010, Technical Report UBLCS-2010-05, University of Bologna.

[20] Sathya, M., M. Swarnamugi, P. Dhavachelvan, and G.

Sureshkumar, Evaluation of qos based web-service selection techniques for service composition. International Journal of Software Engineering, 2010. 1(5): p. 73-90.

[21] Hu, J., C. Guo, H. Wang, and P. Zou. Quality driven web services selection. in e-Business Engineering, 2005. ICEBE

2005. IEEE International Conference. 2005. IEEE.

IJSER © 2013 http://www.ijser.org